374 On the Power of Fluads io conduct Heat. 
conductors of heat, in the way in which solids 
conduct ii, appears to me totally unwarranted 
from. the experiments, and erroneous in itself, 
And as it may be an error of practical conse- 
quence, if adopted, the exposition of it seemed: 
desirable—which,is the object of the following 
remarks. and experiments. 
My first attempt was to ascertain the prectse 
degree of cold \at\ which water ‘ceases to be 
further condensed—and likewise. how much 
expands in cooling below that degree to the 
temperature of freezing, or 32°. For this pur- 
pose I took a thermometer tube, such as would 
have given a scale of 10 inches with mercury 
from 32° to 212°, and filled it with pure water. 
I then graduated it by an accurate mercurial 
thermometer, putting them together into a bason 
filled with water of various degrees of heat, and 
stirring it occasionally : As it is well known, that 
water does not expand in proportion to its heat, 
it does not therefore afford a thermometric scale 
of equal parts, like quicksilver. 
‘From repeated trials agreeing in the result, 
1 find, that the water thermometer is at’ the 
lowest point of the scale it is capable of, that is) 
water is of the greatest density at 42° of the 
mercurial thermometer. From 419% to 44? in- 
clusively the variation is so small as to be just 
perceptible on the scale; but above or below 
